lrrhrrck
md
Lund
IXw~re
/ IYY?)
73.
145.-149
Proposal of an improved score method for the diagnosis of pulmonary tuberculosis in childhood in developing countries G.B. Migliori”, Acocella**
A. Borghesit,
P. Rossanigo 7, C. Adrikot
, M. Neri*, S. SantiniO, A. Bartolonig,
F. Paradisig, G.
* Clinica de1 Lavoro Foundation, Care and Research Institute, Department of Pneumology, Tradate Medical Centre, Tradate, Italy, tDevelopment Cooperation General Direction, Health Section, Ministry of Foreign Affairs, Rome, Italy,* DMO Arua District, Arua, Uganda, 0 Chair of Infectious Diseases, University of Florence, Italy, **Reference Centre for Chemotherapy, University of Pavia, Italy R Y. 210 children aged less than 5 years, referred to the Arua Regional TB Centre (Uganda) for suspected pulmonary tuberculosis (PTB), were examined by anamnesis, clinical examination, Mantoux test, gastric washing, chest X-ray. The response to treatment criterion was applied to the patients treated. According to the score method suggested by Ghidey and Habte, 31 children were diagnosed as PTB patients. 30 of the 31 children with PTB tested positive for alcohol acid-fast bacilli (AAFB) in the aspirated juice. The Mantoux test and X-rays gave a minor contribution to diagnosis. The clinical results are commented. A statistical analysis was carried out to evaluate the role of gastric washing in the diagnosis of PTB in children under 5 years of age (sensitivity, 96.8% ; specificity, 92.2% ; positive predictive value, 68.2 %; negative predictive value, 99.4%). The response to treatment was also evaluated. A modified enlarged score method (based on gastric washing and including response to treatment) is proposed to be applied in developing countries where chest X-ray and other facilities are often lacking. S VMMA
R I?S V MI?. 210 enfants PgQ de mains de 5 ans, soupgonnes d’@treatteints d’une tuberculose pulmonaire (TBP) et orient& vers 1’Arua Regional TB Centre (Uganda), ont Cteexamin& par interrogation, examen clinique, test de Mantoux, lavage gastrique et radiographie thoracique. Le critere de reponse au traitement a CtCapplique aux malades trait&. Selon la methode de cotation suggCrCepar Ghidey et Habte, 31 enfants ont CtCdiagnostiques comme malades TBP. 30 des 31 enfants TBP ont CtCbacteriologiquement confirm& par la presence de bacilles acido-alcoolo-rCsistants (BAAR) dans la liquide de lavage gastrique. Le test de Mantoux et les radiographies ont contribue de falon mineure au diagnostic. Les rCsultats sont comment&. Une analyse statistique a ete effect&e atin d’evaluer le r81e du lavage gastrique dans le diagnostic de la TBP chez les enfants LgCsde moins de 5 ans (sensibilitb 96,8 %; sp&ificitC 92,2 %; valeur predictive positive 68,2 %; valeur predictive negative 99,4%). La reponse au traitement a egalement CtC&al&e. Une methode de cotation modifi& et Clargie (basee sur le lavage gastrique et incluant la reponse au traitement) est propose pour une application dans les pays en developpement oii la radiographie thoracique et d’autres facilites ne sont souvent pas disponibles. Se estudiaron 210 nifios menores de 5 adios de edad, referidos al Centro de Tuberculosis Regional de Arua (Uganda) por sospecha de tuberculosis pulmonar (TBP), mediante una anamnesis, un examen clinico, el test de Mantoux, un lavado gastric0 y una radiograffa de tot-ax. En 10s pacientes tratados se utilize el criteria de respuesta al tratamiento. Segim el metodo de medicion sugerido por Ghidey y Habte, en 31 niiios se hizo el diagnostico de tuberculosis pulmonar. Treinta de ellos presentaban bacilos acid0 alcohol resistentes (BAAR) en el liqido gastric0 aspirado. El test de Mantoux y la radiologia aportaron una minima contribution al diagnbstico. Se comentan 10s resultados clinicos. Se llevo a cabo un analisis estadistico para evaluar el rol de1 lavado gastric0 en el diagndstico de TBP en 10s uifios menores de 5 aiios de edad (sensibilidad R E S VM EN.
Correspondence to: Giovanni Battista Miglioci, Fondazione Lavoro, Via Roncaccio 16/18, 21049 Tradate (VA), Italy.
Clinica del
145
146
Tubercle and Lung Disease
96,8%; especticidad 92,2%; valor predictive positivo 68,2%; valor predictive negativo 99,4%). Tambih se evalub la respuesta al tratamiento. Se propone un m&do de medicih ampliado (basado en el lavado g&trico y que in&ye la respuesta al tratamiento) para ser aplicado en 10s paises en desarrollo, donde a menudo no se dispone de radiografias de t&ax y otros elementos diagnkticos.
INTRODUCTION The diagnosis of pulmonary tuberculosis (PTB) presents special problems in children for several reasons: difficulty in producing sputum; chest X-rays are often negative and not easily available in developing countries; clinical pictures are often atypical; tuberculin tests are difficult to interpret. These problems are even more pronounced in developing countries, where cost-effectiveness aspects must be taken into consideration in the process of selecting the most appropriate diagnostic procedures.‘~* The criteria currently adopted by the Uganda National TB Control Programme for the diagnosis of tuberculosis in childhood are derived from the score method suggested by Ghidey and Habte (Table 1).3 The present study was undertaken with the aim of evaluating the role and contribution of gastric washing technique (GW) and response to treatment (RTT) in a situation with very limited technical facilities in the diagnosis of PTB in children under 5 years of age. Both criteria are simple to apply, cost-effective, and can be performed in a dispensary. In Uganda, where the present study was performed, the Annual Risk of Tuberculosis Infection (ARTI) had been estimated to be 2.6% in 1950,2.3% in 1970 and 2% in 1987, with an estimated incidence of smear positive cases of about 110 per 100 000 population4 The BCG vaccination coverage, evaluated by the authors by means of a baseline survey in 1987 (Expanded Programme of Immunization technique) resulted as follows: l-2 years of age, 51.6%; 5 years of age, 48.8%. Table 1. ref. 3)
Criteria for the diagnosis of tuberculosis
in childhood
(from
1. 2. 3. 4.
History of contact with a tuberculous adult Suggestive symptom complex of TBC Radiological findings compatible with TBC 2 TU PPD reaction positive: 210 mm induration in a non-BCG vaccinated patient 215 mm induration in a BCG vaccinated patient 5. Bacteriological or histological proof Diagnosis The presence of 2 or more criteria was required for diagnosis In the present study the above-mentioned cases only.
criteria were applied to PTB
PATIENTS AND METHODS In the period June 1987 to August 1989, 210 children under 5 years of age were referred for suspected PTB to the Regional Antituberculosis Centre, Arua Hospital, Northern Uganda. Considering the difficulties in evalu-
ating single criteria and diagnosing PTB in children in developing countries, the authors decided to consider as gold standard for PTB the presence of at least 2 criteria as suggested by Ghidey and Habte.3 The criteria adopted were the following: 1. Medical History (MH), sufficient if the child had a household contact with a tuberculous adult. It was collected by a physician during the first accurate clinical examination, considering both personal and family history. Whenever possible the contact was proved by medical records. The BCG scar was carefully checked, together with any physical sign of malnutrition. MH was evaluated in 210 children. 2. Cough (CO) persisting for more than 2 consecutive weeks was considered a symptom suspicious of PTB. Other symptoms (weight loss, fever, sweats), in absence of chronic cough, were considered non specific. CO was evaluated in 210 children. 3. A chest X-ray examination (CXR) was performed on all the children when a consistent suspicion of PTB was present after the physician’s clinical examination. Tomography was not performed. The films were cross-read by the two medical officers in charge of the TB Centre and reviewed by a consulting radiologist. CXR was evaluated in 94 children. 4. A Mantoux test (MAN) with the standard WHO PPD (2 TU of RT 23 Copenhagen containing 0.04 l,tg of PPD from Mycobacterium tuberculosis, 0.01% chinosol and 0.005% tween 80) was performed in the anterior surface of the right forearm and read after 72 h. A result of 10 mm or more induration was considered positive in non-BCG vaccinated children, and 1 of 15 mm in the children with BCG scars. MAN was evaluated in 210 children. In addition 2 other criteria were evaluated: 5. Three nasogastric aspirations were performed early in the morning on an empty stomach, using a Ryle tube 3.3 mm, length 125 mm (Plast AS, DK-3390 Hundested, Denmark), and gastric washing with 2 ml of normal saline. The fluid collected was centrifuged at 4000 rpm for 2 min and the sediment stained according to the Ziehl-Nielsen hot method.5-7 We followed the recommendations suggested by Toman’ for collecting, processing, and staining the specimens. At least 100 immersion fields were read according to the International Union Against Tuberculosis (IUAT) recommendations? also for the quantitative evaluation. The two medical officers in charge of the TB centre cross-read the slides. In case of negative results, 3 new aspirations were performed. In the positive subjects 3 aspirations were
Proposal of an improved
score method for the diagnosis of pulmonary
tuberculosis
in childhood
in developing
countries
147
Table 2. Positivity rate of the different criteria in the group of PTB children (n = 31), in suspected PTB cases (n = 23), and in both groups (n = 54) PTB children 31 (%)
PTB suspected cases 23 (%)
Total 54
(%) 53.7
MH
23
14.2
6
26.1
29
co
24
77.4
9
39.1
33
61.1
CXR
22
70.9
4
17.4
26
48.1
MAN
16
51.6
4
17.4
20
37.0
GW
30
96.7
14
43.5
44
81.5
73.3*
26
66.7$
RTT
15
62.5*
11
*RTT was considered in 24 assessable cases,+RTT was considered $RTT was considered in 39 assessable cases.
done monthly until bacteriological conversion had taken place. 6. In all patients enrolled for treatment body weight was recorded at diagnosis and then monthly to allow evaluation of RTT, defined as clinical improvement (body weight increase superior to 10% of the starting body weight, not inferior to 0.7 kg; cough and nonspecific symptoms disappearing) after 2 months of specific treatment intake. The Uganda TB Control Programme provides as first line treatment streptomycin, thiacetazone, and isoniazid for 2 months followed by thiacetazone and isoniazid for 10 months. 54 children were given treatment. RTT has been evaluated on 39 children who were still alive after 2 months of treatment. Statistical analysis was performed by evaluating sensitivity (SEN), specificity (SPE), positive predictive value (PPV), and negative predictive value (NPV) of GW and RTT in PTB children. Sensitivity was determined as the number of truepositive results divided by the number of true-positive plus false-negative results; specificity was the number of true-negative results divided by the number of truenegative plus false-positive results. The predictive value of a positive test was calculated as the number of the positive results divided by all true- and false- positive results. The predictive value of a negative test was calculated as the number of true-negative results divided by all true- and false-negative results. In addition PTB diagnosis rate in a hypothetical dispensary (without CXR facilities and fridge to store PPD) is calculated with the aim of evaluating the association of PTB cases with positive GW and/or RTT (using only 2 criteria, namely CO and MH). Particular individual attention will be paid to the patients with positive RTT criterion and repeated GW examinations after 2 months of specific treatment intake.
RESULTS According to Ghidey and Habte’s score method we may divide the children into 3 groups:
in 15 assessable cases,
(4 PTB cases (31 children): more than 2 points were
(B)
(C)
collected with or without GW and RTT criteria satisfied; Highly suspected PTB cases (23 children): 1 point was collected; in addition, positive GW and/or RTT criteria were satisfied (they were enrolled for specific treatment); Suspected PTB cases (156 children): 1 point was collected; GW was negative. RTT was not evaluated because they were not enrolled for specific treatment.
The positivity rate of the different criteria is shown in Table 2. In group A (PTB cases) 24 children had typical symptoms, presenting chronic cough plus minor signs. 23 patients (74.2%) lived in a household with a person affected by PTB; 9 of them (39.1%) had BCG scar. Among PTB children, 16 (51.6%) had a positive tuberculin test, 7 having BCG scar (43.7%). As far as CXR is concerned, in group A, 22 patients (70.9%) had CXR suggestive of PTB; 7 cases presented hilar and mediastinal lymphadenopathy, 6 cavitarian lesions (1 of them together with a miliary picture), 8 segmental infiltrations and I a nodular apical infiltration.’ Four PTB compatible films (2 hilar and mediastinal lymphadenopathy and 2 segmental infiltration) were found among the group B patients. Of the 40 patients of group C controlled by CXR, 13 had TB compatible-film, but all of them improved after aspecific antibiotic treatment. Among PTB cases we found 8 children with non-compatible CXR and positive GW (25.8%); 7 of them (87.5%) did not present gastric juice conversion from positive to negative until after a Table 3. Results of gastric washing in 210 children referred for suspected PTB Children with at least 1 slide positive for AAPES = 44 with 3 slides positive out of 3 = 16 with 2 slides positive out of 3 = 12 with 1 slide positive out of 3 = 15 I with 1 slide positive out of 6 = with 3+ (> 10 AAFB per field) = 3 with 2+ (l-10 AAFB per field) = 6 with 1+ (< 99 AAFB per 100 immersion fields) = 35
-
148
Tubercle and Lung Disease
minimum of 2 months of treatment, and 7 had a positive RTT (1 was not assessable). Out of 210 children, 44 had at least 1 gastric aspirate juice positive for AAFB (Table 3), 30 belonging to PTB cases and 14 to highly suspected PTB cases. The RTT criterion was evaluated in 24 patients. The average body weight at diagnosis among PTB cases was 9 kg (range 3.6-23 kg). The criterion was positive in 15 (62.5%). The average body weight increase after 2 months of treatment was 1 kg, ranging from O-5.5 kg. As regards the follow-up, 14 of the 54 patients treated improved (25.9%) with an average increase of 5.2 kg (ranging 0.2-13 kg), 11 died (20.4%), 17 defaulted (31.5%), and 12 (22.2%) still continue treatment regularly. Out of the 14 patients that improved, E belong to PTB cases and 6 are highly suspected TB cases. Bacteriological conversion was evaluated in 31 patients (23 from group A). 18 (58.1%) converted after 2 months, 8 (25.8%) after 3 months and 5 (16.1%) after 4 months. SEN, SPE, PPV, and NPV of GW and BTT are shown in Table 4. In both tests SPE and NPV are high (superior to 90%). GW has also a very high sensitivity (96.8%), RTT having a value of 62.5%. Less indicative appears to be PPV (GW = 68.2%, RTT = 57.7%). In Table 5 we have simulated the diagnostic rate at hospital level (using 4 criteria: MH+CO+CXR+MAN) and in a hypothetical dispensary lacking a fridge and X-ray facilities (using MH+CO). At hospital level we obtained 31 diagnoses (13 children having 2 points, 13 children 3 points, and 5 children 4 points), while at dispensary level 16 (2 children having 2 points, 9 children 3 points, and 5 children 4 points). The diagnostic rate has lowered from 100% to 51.6% applying Ghidey and Habte’ s score. The good specificity and sensitivity values of GW and RTT encouraged us to utilize these 2 criteria in an enlarged score (MH+CO+CXR+MAN+GW+RTT) at Table 4. Sensitivity (SEN), specificity (SPE), positive (PPV) and negative (NPV) predictive values of gastric washing (GW) and response to treatment (RTT) criteria in 31 PTB children. GW GW+ GWTotal
PTB+ 30 1 31
PTB14 165 179
Total 44 166 210
SEN SPE PPV NPV
= = = =
96.8% 92.2% 68.2% 99.4%
SEN SPE PPV NPV
= = = =
62.5% 94.1% 57.7% 95.1%
RTT RTT+ RTTTotal
PTB+ 15 9 24
PTB11 175 186
Total 26 184 210
hospital and dispensary level (right column of Table 5). The same diagnostic rate (31 diagnoses of PTB) was obtained considering or not MAN and CXR. In Table 3 the quantitative evaluation of positive GW performed is summarized. Among 31 PTB children, 23 (74.2%) had repeated positive GW after the second month of specific treatment. RTT was positive in 14 of them (60.9%). Among 23 highly suspected PTB cases, 8 (34.8%) had repeated positive GW, with 5 positive RTT (62.5%).
DISCUSSION An examination of sputum is the most important method for diagnosis of PTB in developing countries.” A smear of gastric aspirates is considered an acceptable substitute when sputum is not produced,“~‘* and therefore is particularly suitable in young children. On the contrary, X-ray examination plays a minor role in PTB detection because of the high cost and the high rate of falsepositive results due to its low specificity.‘3,‘4 In our study the gastric washing technique gave satisfactory results. In particular, it was well accepted by staff and mothers, it allowed us to make an accurate diagnosis in most cases, with satisfactory sensitivity and specificity values, it can be used to monitor the patient’s response to treatment if it is performed monthly until juice conversion takes place, and it is cost-effective. In a developing country with a high prevalence of TB the probability of a false-positive gastric aspiration due to acid-fast saprophytes in the stomach in a child with chest infection and positive to a diagnostic score method for TB appears to be very 10w.l~ Moreover, this falsepositive result could be explained by the fact that during treatment for proved tuberculosis patients may excrete AAFB which do not grow in culture.” In our study 30 out of 3 1 children diagnosed as PTB had a positive GW. The association between PTB disease and the presence of AAFB in the stomach appears to be important. We noted a discrepancy between CXR and GW positivity. We can rule out the possibility of false-positive results due to the presence of ubiquitous acid-fast saprophytes in the majority of the cases, considering that 8 out of 31 PTB cases had negative CXR with positive GW, but 7 of them (87.5%) had repeated positive GW and positive RTT. The low sensitivity of the X-rays could be related to the well known difficulties in obtaining good radiographic pictures in young children (lung hypoinflation, non-rotated chest radiograms), as well as to the
Table 5. Comparison of the diagnostic rate (%) of the score suggested by Ghidey and Habte (G & H) and of the enlarged score considering complete and reduced application of the score (+GW+RTT) Criteria used
Score G & H
Enlarged score (+GW+RTT)
MH+CO+CXR+MAN
31(100%)
31 (100%)
MH+CO
16 (51.6%)
31(100%)
Proposal of an improved score method for the diagnosis
unavailability of tomography. Our findings indicate that smears of gastric washing are much more useful than X-rays for the purpose of diagnosis of PTB, at least in paediatric patients. Moreover, other investigators16 found similar discrepancies in adults where chest X-ray interpretation appears to be easier. We are aware of the main limits of our study: 1. the impossibility of confirming positive GW with culture for lack of facilities (a common problem in developing countries): 2. the impossibility of having, as control, a group of PTB children diagnosed without GW and RTT, but applying Ghidey and Habte’s score. We cannot use Ghidey’s paper as an historical control because it does not report results on negative children and is applied to all TB forms. Anyway, we may assume that in developing countries with high prevalence of TB, a child positive to a score with AAFB in the stomach really has PTB. In the same way, we have a low probability that a child with AAFB in the stomach and/or RTT satisfied does not have PTB disease, particularly if another criterion is positive. On those assumptions, the authors suggest a new gastric washing based score method (Table 6), to overcome some of the limitations of Ghidey and Habte’s scoring which: 1. is less specific for PTB (representing the majority of TB cases in subjects under 5 years of age); 2. doesn’t explain how to collect material for the bacteriological proof and does not consider that a positive bacilloscopy alone is sufficient for a correct diagnosis; 3. requires facilities (e.g. radiology) available only at hospitals; 4. does not include the RTT criterion, which can help in is particular cases (e.g. when another criterion required for diagnosis). The new score method we propose can be used both in dispensary situations (using a simplified procedure including MH, CO, and GW) and hospital. Sensitivity and PPV of GW and RTT are probably even higher if extended to the population of highly suspected TB cases. Our study indicates that the simplified procedure does not reduce the diagnostic accuracy in a group of 3 1 Table 6. Proposed criteria for the diagnosis of pulmonary tuberculosis in childhood (A) Gastric washing positive for AAFB or (B) Two or more of the following criteria History of contact with a tuberculous adult Suggestive symptoms of PTB (cough for more. than 2 weeks) 2 TU PPD reaction positive _>10 mm in unvaccinated BCG patients 2 1.5mm in vaccinated BCG patients Radiological findings compatible with PTB Response to treatment (body weight increased > 10% after 2 months of treatment, plus clinical improvement)
of pulmonary
tuberculosis
in childhood
in developing
countries
149
PTB patients. However, in developing countries, where culture facilities are usually not available, gastric washing and direct microscopy can be used for routine diagnosis of TB in childhood just as gastric lavage, sputum and bronchial washing are used for TB culturing in developed countries.‘7 We hope that our experience may contribute to improve the diagnostic standards of PTB in developing countries, minimizing unnecessary X-ray examinations and lowering the cost of each diagnosis.” Further studies on the same topic are encouraged.
Acknowledgements The Authors wish to thank Ing. Luigi Ballardini and Emanuela Radice (Statistical Department, Clinica de1 Lavoro Foundation, Tradate) for their considerable help in the data analysis.
References 1. Amodio J. Abramson S, Berdon W. Primary pulmonary tuberculosis in infancy: a resurgent disease in the urban United States. Pediatr Radio1 1986; 16: 185-189. 2. Anane T. The problem of child tuberculosis and its magnitude in Algerie. Proceedings of the 26th IUAT World Conference on Tuberculosis, Singapore, 1986: Abstract AO13. 3. Ghidey Y, Habte D. Tuberculosis in childhood: an analysis of 4 12 cases. Ethiop Med J 1983; 21: 161-167. 4. Migliori G B, Borghesi A, Adriko C, Santini S, Spanevello A, Acocella G. Bias obtained by selecting schools as clusters during a tuberculin survey in North Western Uganda. Am Rev Respir Dis 1990; 141: A802. 5. Ebrahim G J. Paediatric Practice in Developing Countries. EL.BS., London, 1984. confirmed 6. Donald P R, Ball J B, Burger P J. Bacteriologically pulmonary tuberculosis in childhood. Clinical and radiological features. S Afr Med J 1985; 67: 588-590. 7. Biersack G. Examination of sputum for acid-fast bacilli by the concentration method (a reminder). Trop Doct 1988; 18: 53. 8. Toman K. Tuberculosis case-finding and chemotherapy. Questions and answers. Geneva WHO: 1979. 9. IUAT. Technical guide, for sputum examination for tuberculosis by direct microscopy. International Union Against Tuberculosis, 1978. 10. Bell D R. Lecture notes on tropical medicine, 2nd ed. Oxford Blackwell : 1985. 11. Daniel T M. Tuberculosis. In: Warren K S, Mahmoud A A F, eds. Tropical and geographical medicine. New York: McGraw Hill, 1984: pp 786-797. 12. Pratt P C, Atwell R J. The diagnostic reliability of acid-fast bacilli demonstrated in aspirated gastric contents. Am Rev Respir Dis 1961; 83: 96-99. 13. Chaulet P, Mulder D. Tuberculosis. In: Manson Bahr P E C, Bell D R, eds. Manson’s tropical diseases, 19th ed. London: Bailliere Tindall, 1987: pp 987-997. 14. Gordin F M, Slutkin G, Schecter G, Goodman P C, Hopewell P C. Presumptive diagnosis and treatment of pulmonary tuberculosis based on radiographic findings. Am Rev Respir Dis 1989; 139: 1090-1093. 15. Migliori G B, Borghesi A, Oryem V, Comaschi E, Vocaturo G, Neri M. Gastric washing to evaluate the clinical outcome of under 5 children affected by Pulmonary Tuberculosis. Abstract: Joint Meeting SEP-SEPCR, London, 1990. 16. Nyboe J. Results of the international study on x-ray classification. Bull Int Union Tuberc Lung Dis 1968,41, 115-124. 17. Olofsson J. Comparison of gastric lavage, sputum and bronchial washing for tuberculosis culturing. Proceedings of the 26th IUAT World Conference on Tuberculosis, Singapore, 1986: Abstract A232. 18. Whittaker L R. Minimizing unnecessary X-ray examinations. A national and professional approach. Trop Doct 1987; 17: 62-66.
md
Lund
IXw~re
/ IYY?)
73.
145.-149
Proposal of an improved score method for the diagnosis of pulmonary tuberculosis in childhood in developing countries G.B. Migliori”, Acocella**
A. Borghesit,
P. Rossanigo 7, C. Adrikot
, M. Neri*, S. SantiniO, A. Bartolonig,
F. Paradisig, G.
* Clinica de1 Lavoro Foundation, Care and Research Institute, Department of Pneumology, Tradate Medical Centre, Tradate, Italy, tDevelopment Cooperation General Direction, Health Section, Ministry of Foreign Affairs, Rome, Italy,* DMO Arua District, Arua, Uganda, 0 Chair of Infectious Diseases, University of Florence, Italy, **Reference Centre for Chemotherapy, University of Pavia, Italy R Y. 210 children aged less than 5 years, referred to the Arua Regional TB Centre (Uganda) for suspected pulmonary tuberculosis (PTB), were examined by anamnesis, clinical examination, Mantoux test, gastric washing, chest X-ray. The response to treatment criterion was applied to the patients treated. According to the score method suggested by Ghidey and Habte, 31 children were diagnosed as PTB patients. 30 of the 31 children with PTB tested positive for alcohol acid-fast bacilli (AAFB) in the aspirated juice. The Mantoux test and X-rays gave a minor contribution to diagnosis. The clinical results are commented. A statistical analysis was carried out to evaluate the role of gastric washing in the diagnosis of PTB in children under 5 years of age (sensitivity, 96.8% ; specificity, 92.2% ; positive predictive value, 68.2 %; negative predictive value, 99.4%). The response to treatment was also evaluated. A modified enlarged score method (based on gastric washing and including response to treatment) is proposed to be applied in developing countries where chest X-ray and other facilities are often lacking. S VMMA
R I?S V MI?. 210 enfants PgQ de mains de 5 ans, soupgonnes d’@treatteints d’une tuberculose pulmonaire (TBP) et orient& vers 1’Arua Regional TB Centre (Uganda), ont Cteexamin& par interrogation, examen clinique, test de Mantoux, lavage gastrique et radiographie thoracique. Le critere de reponse au traitement a CtCapplique aux malades trait&. Selon la methode de cotation suggCrCepar Ghidey et Habte, 31 enfants ont CtCdiagnostiques comme malades TBP. 30 des 31 enfants TBP ont CtCbacteriologiquement confirm& par la presence de bacilles acido-alcoolo-rCsistants (BAAR) dans la liquide de lavage gastrique. Le test de Mantoux et les radiographies ont contribue de falon mineure au diagnostic. Les rCsultats sont comment&. Une analyse statistique a ete effect&e atin d’evaluer le r81e du lavage gastrique dans le diagnostic de la TBP chez les enfants LgCsde moins de 5 ans (sensibilitb 96,8 %; sp&ificitC 92,2 %; valeur predictive positive 68,2 %; valeur predictive negative 99,4%). La reponse au traitement a egalement CtC&al&e. Une methode de cotation modifi& et Clargie (basee sur le lavage gastrique et incluant la reponse au traitement) est propose pour une application dans les pays en developpement oii la radiographie thoracique et d’autres facilites ne sont souvent pas disponibles. Se estudiaron 210 nifios menores de 5 adios de edad, referidos al Centro de Tuberculosis Regional de Arua (Uganda) por sospecha de tuberculosis pulmonar (TBP), mediante una anamnesis, un examen clinico, el test de Mantoux, un lavado gastric0 y una radiograffa de tot-ax. En 10s pacientes tratados se utilize el criteria de respuesta al tratamiento. Segim el metodo de medicion sugerido por Ghidey y Habte, en 31 niiios se hizo el diagnostico de tuberculosis pulmonar. Treinta de ellos presentaban bacilos acid0 alcohol resistentes (BAAR) en el liqido gastric0 aspirado. El test de Mantoux y la radiologia aportaron una minima contribution al diagnbstico. Se comentan 10s resultados clinicos. Se llevo a cabo un analisis estadistico para evaluar el rol de1 lavado gastric0 en el diagndstico de TBP en 10s uifios menores de 5 aiios de edad (sensibilidad R E S VM EN.
Correspondence to: Giovanni Battista Miglioci, Fondazione Lavoro, Via Roncaccio 16/18, 21049 Tradate (VA), Italy.
Clinica del
145
146
Tubercle and Lung Disease
96,8%; especticidad 92,2%; valor predictive positivo 68,2%; valor predictive negativo 99,4%). Tambih se evalub la respuesta al tratamiento. Se propone un m&do de medicih ampliado (basado en el lavado g&trico y que in&ye la respuesta al tratamiento) para ser aplicado en 10s paises en desarrollo, donde a menudo no se dispone de radiografias de t&ax y otros elementos diagnkticos.
INTRODUCTION The diagnosis of pulmonary tuberculosis (PTB) presents special problems in children for several reasons: difficulty in producing sputum; chest X-rays are often negative and not easily available in developing countries; clinical pictures are often atypical; tuberculin tests are difficult to interpret. These problems are even more pronounced in developing countries, where cost-effectiveness aspects must be taken into consideration in the process of selecting the most appropriate diagnostic procedures.‘~* The criteria currently adopted by the Uganda National TB Control Programme for the diagnosis of tuberculosis in childhood are derived from the score method suggested by Ghidey and Habte (Table 1).3 The present study was undertaken with the aim of evaluating the role and contribution of gastric washing technique (GW) and response to treatment (RTT) in a situation with very limited technical facilities in the diagnosis of PTB in children under 5 years of age. Both criteria are simple to apply, cost-effective, and can be performed in a dispensary. In Uganda, where the present study was performed, the Annual Risk of Tuberculosis Infection (ARTI) had been estimated to be 2.6% in 1950,2.3% in 1970 and 2% in 1987, with an estimated incidence of smear positive cases of about 110 per 100 000 population4 The BCG vaccination coverage, evaluated by the authors by means of a baseline survey in 1987 (Expanded Programme of Immunization technique) resulted as follows: l-2 years of age, 51.6%; 5 years of age, 48.8%. Table 1. ref. 3)
Criteria for the diagnosis of tuberculosis
in childhood
(from
1. 2. 3. 4.
History of contact with a tuberculous adult Suggestive symptom complex of TBC Radiological findings compatible with TBC 2 TU PPD reaction positive: 210 mm induration in a non-BCG vaccinated patient 215 mm induration in a BCG vaccinated patient 5. Bacteriological or histological proof Diagnosis The presence of 2 or more criteria was required for diagnosis In the present study the above-mentioned cases only.
criteria were applied to PTB
PATIENTS AND METHODS In the period June 1987 to August 1989, 210 children under 5 years of age were referred for suspected PTB to the Regional Antituberculosis Centre, Arua Hospital, Northern Uganda. Considering the difficulties in evalu-
ating single criteria and diagnosing PTB in children in developing countries, the authors decided to consider as gold standard for PTB the presence of at least 2 criteria as suggested by Ghidey and Habte.3 The criteria adopted were the following: 1. Medical History (MH), sufficient if the child had a household contact with a tuberculous adult. It was collected by a physician during the first accurate clinical examination, considering both personal and family history. Whenever possible the contact was proved by medical records. The BCG scar was carefully checked, together with any physical sign of malnutrition. MH was evaluated in 210 children. 2. Cough (CO) persisting for more than 2 consecutive weeks was considered a symptom suspicious of PTB. Other symptoms (weight loss, fever, sweats), in absence of chronic cough, were considered non specific. CO was evaluated in 210 children. 3. A chest X-ray examination (CXR) was performed on all the children when a consistent suspicion of PTB was present after the physician’s clinical examination. Tomography was not performed. The films were cross-read by the two medical officers in charge of the TB Centre and reviewed by a consulting radiologist. CXR was evaluated in 94 children. 4. A Mantoux test (MAN) with the standard WHO PPD (2 TU of RT 23 Copenhagen containing 0.04 l,tg of PPD from Mycobacterium tuberculosis, 0.01% chinosol and 0.005% tween 80) was performed in the anterior surface of the right forearm and read after 72 h. A result of 10 mm or more induration was considered positive in non-BCG vaccinated children, and 1 of 15 mm in the children with BCG scars. MAN was evaluated in 210 children. In addition 2 other criteria were evaluated: 5. Three nasogastric aspirations were performed early in the morning on an empty stomach, using a Ryle tube 3.3 mm, length 125 mm (Plast AS, DK-3390 Hundested, Denmark), and gastric washing with 2 ml of normal saline. The fluid collected was centrifuged at 4000 rpm for 2 min and the sediment stained according to the Ziehl-Nielsen hot method.5-7 We followed the recommendations suggested by Toman’ for collecting, processing, and staining the specimens. At least 100 immersion fields were read according to the International Union Against Tuberculosis (IUAT) recommendations? also for the quantitative evaluation. The two medical officers in charge of the TB centre cross-read the slides. In case of negative results, 3 new aspirations were performed. In the positive subjects 3 aspirations were
Proposal of an improved
score method for the diagnosis of pulmonary
tuberculosis
in childhood
in developing
countries
147
Table 2. Positivity rate of the different criteria in the group of PTB children (n = 31), in suspected PTB cases (n = 23), and in both groups (n = 54) PTB children 31 (%)
PTB suspected cases 23 (%)
Total 54
(%) 53.7
MH
23
14.2
6
26.1
29
co
24
77.4
9
39.1
33
61.1
CXR
22
70.9
4
17.4
26
48.1
MAN
16
51.6
4
17.4
20
37.0
GW
30
96.7
14
43.5
44
81.5
73.3*
26
66.7$
RTT
15
62.5*
11
*RTT was considered in 24 assessable cases,+RTT was considered $RTT was considered in 39 assessable cases.
done monthly until bacteriological conversion had taken place. 6. In all patients enrolled for treatment body weight was recorded at diagnosis and then monthly to allow evaluation of RTT, defined as clinical improvement (body weight increase superior to 10% of the starting body weight, not inferior to 0.7 kg; cough and nonspecific symptoms disappearing) after 2 months of specific treatment intake. The Uganda TB Control Programme provides as first line treatment streptomycin, thiacetazone, and isoniazid for 2 months followed by thiacetazone and isoniazid for 10 months. 54 children were given treatment. RTT has been evaluated on 39 children who were still alive after 2 months of treatment. Statistical analysis was performed by evaluating sensitivity (SEN), specificity (SPE), positive predictive value (PPV), and negative predictive value (NPV) of GW and RTT in PTB children. Sensitivity was determined as the number of truepositive results divided by the number of true-positive plus false-negative results; specificity was the number of true-negative results divided by the number of truenegative plus false-positive results. The predictive value of a positive test was calculated as the number of the positive results divided by all true- and false- positive results. The predictive value of a negative test was calculated as the number of true-negative results divided by all true- and false-negative results. In addition PTB diagnosis rate in a hypothetical dispensary (without CXR facilities and fridge to store PPD) is calculated with the aim of evaluating the association of PTB cases with positive GW and/or RTT (using only 2 criteria, namely CO and MH). Particular individual attention will be paid to the patients with positive RTT criterion and repeated GW examinations after 2 months of specific treatment intake.
RESULTS According to Ghidey and Habte’s score method we may divide the children into 3 groups:
in 15 assessable cases,
(4 PTB cases (31 children): more than 2 points were
(B)
(C)
collected with or without GW and RTT criteria satisfied; Highly suspected PTB cases (23 children): 1 point was collected; in addition, positive GW and/or RTT criteria were satisfied (they were enrolled for specific treatment); Suspected PTB cases (156 children): 1 point was collected; GW was negative. RTT was not evaluated because they were not enrolled for specific treatment.
The positivity rate of the different criteria is shown in Table 2. In group A (PTB cases) 24 children had typical symptoms, presenting chronic cough plus minor signs. 23 patients (74.2%) lived in a household with a person affected by PTB; 9 of them (39.1%) had BCG scar. Among PTB children, 16 (51.6%) had a positive tuberculin test, 7 having BCG scar (43.7%). As far as CXR is concerned, in group A, 22 patients (70.9%) had CXR suggestive of PTB; 7 cases presented hilar and mediastinal lymphadenopathy, 6 cavitarian lesions (1 of them together with a miliary picture), 8 segmental infiltrations and I a nodular apical infiltration.’ Four PTB compatible films (2 hilar and mediastinal lymphadenopathy and 2 segmental infiltration) were found among the group B patients. Of the 40 patients of group C controlled by CXR, 13 had TB compatible-film, but all of them improved after aspecific antibiotic treatment. Among PTB cases we found 8 children with non-compatible CXR and positive GW (25.8%); 7 of them (87.5%) did not present gastric juice conversion from positive to negative until after a Table 3. Results of gastric washing in 210 children referred for suspected PTB Children with at least 1 slide positive for AAPES = 44 with 3 slides positive out of 3 = 16 with 2 slides positive out of 3 = 12 with 1 slide positive out of 3 = 15 I with 1 slide positive out of 6 = with 3+ (> 10 AAFB per field) = 3 with 2+ (l-10 AAFB per field) = 6 with 1+ (< 99 AAFB per 100 immersion fields) = 35
-
148
Tubercle and Lung Disease
minimum of 2 months of treatment, and 7 had a positive RTT (1 was not assessable). Out of 210 children, 44 had at least 1 gastric aspirate juice positive for AAFB (Table 3), 30 belonging to PTB cases and 14 to highly suspected PTB cases. The RTT criterion was evaluated in 24 patients. The average body weight at diagnosis among PTB cases was 9 kg (range 3.6-23 kg). The criterion was positive in 15 (62.5%). The average body weight increase after 2 months of treatment was 1 kg, ranging from O-5.5 kg. As regards the follow-up, 14 of the 54 patients treated improved (25.9%) with an average increase of 5.2 kg (ranging 0.2-13 kg), 11 died (20.4%), 17 defaulted (31.5%), and 12 (22.2%) still continue treatment regularly. Out of the 14 patients that improved, E belong to PTB cases and 6 are highly suspected TB cases. Bacteriological conversion was evaluated in 31 patients (23 from group A). 18 (58.1%) converted after 2 months, 8 (25.8%) after 3 months and 5 (16.1%) after 4 months. SEN, SPE, PPV, and NPV of GW and BTT are shown in Table 4. In both tests SPE and NPV are high (superior to 90%). GW has also a very high sensitivity (96.8%), RTT having a value of 62.5%. Less indicative appears to be PPV (GW = 68.2%, RTT = 57.7%). In Table 5 we have simulated the diagnostic rate at hospital level (using 4 criteria: MH+CO+CXR+MAN) and in a hypothetical dispensary lacking a fridge and X-ray facilities (using MH+CO). At hospital level we obtained 31 diagnoses (13 children having 2 points, 13 children 3 points, and 5 children 4 points), while at dispensary level 16 (2 children having 2 points, 9 children 3 points, and 5 children 4 points). The diagnostic rate has lowered from 100% to 51.6% applying Ghidey and Habte’ s score. The good specificity and sensitivity values of GW and RTT encouraged us to utilize these 2 criteria in an enlarged score (MH+CO+CXR+MAN+GW+RTT) at Table 4. Sensitivity (SEN), specificity (SPE), positive (PPV) and negative (NPV) predictive values of gastric washing (GW) and response to treatment (RTT) criteria in 31 PTB children. GW GW+ GWTotal
PTB+ 30 1 31
PTB14 165 179
Total 44 166 210
SEN SPE PPV NPV
= = = =
96.8% 92.2% 68.2% 99.4%
SEN SPE PPV NPV
= = = =
62.5% 94.1% 57.7% 95.1%
RTT RTT+ RTTTotal
PTB+ 15 9 24
PTB11 175 186
Total 26 184 210
hospital and dispensary level (right column of Table 5). The same diagnostic rate (31 diagnoses of PTB) was obtained considering or not MAN and CXR. In Table 3 the quantitative evaluation of positive GW performed is summarized. Among 31 PTB children, 23 (74.2%) had repeated positive GW after the second month of specific treatment. RTT was positive in 14 of them (60.9%). Among 23 highly suspected PTB cases, 8 (34.8%) had repeated positive GW, with 5 positive RTT (62.5%).
DISCUSSION An examination of sputum is the most important method for diagnosis of PTB in developing countries.” A smear of gastric aspirates is considered an acceptable substitute when sputum is not produced,“~‘* and therefore is particularly suitable in young children. On the contrary, X-ray examination plays a minor role in PTB detection because of the high cost and the high rate of falsepositive results due to its low specificity.‘3,‘4 In our study the gastric washing technique gave satisfactory results. In particular, it was well accepted by staff and mothers, it allowed us to make an accurate diagnosis in most cases, with satisfactory sensitivity and specificity values, it can be used to monitor the patient’s response to treatment if it is performed monthly until juice conversion takes place, and it is cost-effective. In a developing country with a high prevalence of TB the probability of a false-positive gastric aspiration due to acid-fast saprophytes in the stomach in a child with chest infection and positive to a diagnostic score method for TB appears to be very 10w.l~ Moreover, this falsepositive result could be explained by the fact that during treatment for proved tuberculosis patients may excrete AAFB which do not grow in culture.” In our study 30 out of 3 1 children diagnosed as PTB had a positive GW. The association between PTB disease and the presence of AAFB in the stomach appears to be important. We noted a discrepancy between CXR and GW positivity. We can rule out the possibility of false-positive results due to the presence of ubiquitous acid-fast saprophytes in the majority of the cases, considering that 8 out of 31 PTB cases had negative CXR with positive GW, but 7 of them (87.5%) had repeated positive GW and positive RTT. The low sensitivity of the X-rays could be related to the well known difficulties in obtaining good radiographic pictures in young children (lung hypoinflation, non-rotated chest radiograms), as well as to the
Table 5. Comparison of the diagnostic rate (%) of the score suggested by Ghidey and Habte (G & H) and of the enlarged score considering complete and reduced application of the score (+GW+RTT) Criteria used
Score G & H
Enlarged score (+GW+RTT)
MH+CO+CXR+MAN
31(100%)
31 (100%)
MH+CO
16 (51.6%)
31(100%)
Proposal of an improved score method for the diagnosis
unavailability of tomography. Our findings indicate that smears of gastric washing are much more useful than X-rays for the purpose of diagnosis of PTB, at least in paediatric patients. Moreover, other investigators16 found similar discrepancies in adults where chest X-ray interpretation appears to be easier. We are aware of the main limits of our study: 1. the impossibility of confirming positive GW with culture for lack of facilities (a common problem in developing countries): 2. the impossibility of having, as control, a group of PTB children diagnosed without GW and RTT, but applying Ghidey and Habte’s score. We cannot use Ghidey’s paper as an historical control because it does not report results on negative children and is applied to all TB forms. Anyway, we may assume that in developing countries with high prevalence of TB, a child positive to a score with AAFB in the stomach really has PTB. In the same way, we have a low probability that a child with AAFB in the stomach and/or RTT satisfied does not have PTB disease, particularly if another criterion is positive. On those assumptions, the authors suggest a new gastric washing based score method (Table 6), to overcome some of the limitations of Ghidey and Habte’s scoring which: 1. is less specific for PTB (representing the majority of TB cases in subjects under 5 years of age); 2. doesn’t explain how to collect material for the bacteriological proof and does not consider that a positive bacilloscopy alone is sufficient for a correct diagnosis; 3. requires facilities (e.g. radiology) available only at hospitals; 4. does not include the RTT criterion, which can help in is particular cases (e.g. when another criterion required for diagnosis). The new score method we propose can be used both in dispensary situations (using a simplified procedure including MH, CO, and GW) and hospital. Sensitivity and PPV of GW and RTT are probably even higher if extended to the population of highly suspected TB cases. Our study indicates that the simplified procedure does not reduce the diagnostic accuracy in a group of 3 1 Table 6. Proposed criteria for the diagnosis of pulmonary tuberculosis in childhood (A) Gastric washing positive for AAFB or (B) Two or more of the following criteria History of contact with a tuberculous adult Suggestive symptoms of PTB (cough for more. than 2 weeks) 2 TU PPD reaction positive _>10 mm in unvaccinated BCG patients 2 1.5mm in vaccinated BCG patients Radiological findings compatible with PTB Response to treatment (body weight increased > 10% after 2 months of treatment, plus clinical improvement)
of pulmonary
tuberculosis
in childhood
in developing
countries
149
PTB patients. However, in developing countries, where culture facilities are usually not available, gastric washing and direct microscopy can be used for routine diagnosis of TB in childhood just as gastric lavage, sputum and bronchial washing are used for TB culturing in developed countries.‘7 We hope that our experience may contribute to improve the diagnostic standards of PTB in developing countries, minimizing unnecessary X-ray examinations and lowering the cost of each diagnosis.” Further studies on the same topic are encouraged.
Acknowledgements The Authors wish to thank Ing. Luigi Ballardini and Emanuela Radice (Statistical Department, Clinica de1 Lavoro Foundation, Tradate) for their considerable help in the data analysis.
References 1. Amodio J. Abramson S, Berdon W. Primary pulmonary tuberculosis in infancy: a resurgent disease in the urban United States. Pediatr Radio1 1986; 16: 185-189. 2. Anane T. The problem of child tuberculosis and its magnitude in Algerie. Proceedings of the 26th IUAT World Conference on Tuberculosis, Singapore, 1986: Abstract AO13. 3. Ghidey Y, Habte D. Tuberculosis in childhood: an analysis of 4 12 cases. Ethiop Med J 1983; 21: 161-167. 4. Migliori G B, Borghesi A, Adriko C, Santini S, Spanevello A, Acocella G. Bias obtained by selecting schools as clusters during a tuberculin survey in North Western Uganda. Am Rev Respir Dis 1990; 141: A802. 5. Ebrahim G J. Paediatric Practice in Developing Countries. EL.BS., London, 1984. confirmed 6. Donald P R, Ball J B, Burger P J. Bacteriologically pulmonary tuberculosis in childhood. Clinical and radiological features. S Afr Med J 1985; 67: 588-590. 7. Biersack G. Examination of sputum for acid-fast bacilli by the concentration method (a reminder). Trop Doct 1988; 18: 53. 8. Toman K. Tuberculosis case-finding and chemotherapy. Questions and answers. Geneva WHO: 1979. 9. IUAT. Technical guide, for sputum examination for tuberculosis by direct microscopy. International Union Against Tuberculosis, 1978. 10. Bell D R. Lecture notes on tropical medicine, 2nd ed. Oxford Blackwell : 1985. 11. Daniel T M. Tuberculosis. In: Warren K S, Mahmoud A A F, eds. Tropical and geographical medicine. New York: McGraw Hill, 1984: pp 786-797. 12. Pratt P C, Atwell R J. The diagnostic reliability of acid-fast bacilli demonstrated in aspirated gastric contents. Am Rev Respir Dis 1961; 83: 96-99. 13. Chaulet P, Mulder D. Tuberculosis. In: Manson Bahr P E C, Bell D R, eds. Manson’s tropical diseases, 19th ed. London: Bailliere Tindall, 1987: pp 987-997. 14. Gordin F M, Slutkin G, Schecter G, Goodman P C, Hopewell P C. Presumptive diagnosis and treatment of pulmonary tuberculosis based on radiographic findings. Am Rev Respir Dis 1989; 139: 1090-1093. 15. Migliori G B, Borghesi A, Oryem V, Comaschi E, Vocaturo G, Neri M. Gastric washing to evaluate the clinical outcome of under 5 children affected by Pulmonary Tuberculosis. Abstract: Joint Meeting SEP-SEPCR, London, 1990. 16. Nyboe J. Results of the international study on x-ray classification. Bull Int Union Tuberc Lung Dis 1968,41, 115-124. 17. Olofsson J. Comparison of gastric lavage, sputum and bronchial washing for tuberculosis culturing. Proceedings of the 26th IUAT World Conference on Tuberculosis, Singapore, 1986: Abstract A232. 18. Whittaker L R. Minimizing unnecessary X-ray examinations. A national and professional approach. Trop Doct 1987; 17: 62-66.
